Synthesis and Spectral and Chemical Properties of the Yellow

Jul 18, 2009 - pathway based on autocatalytic modifications of their amino acid residues. The yellow fluorescent protein. zFP538 from the button polyp...
1 downloads 0 Views 947KB Size
Biochemistry 2009, 48, 8077–8082 8077 DOI: 10.1021/bi900719x

Synthesis and Spectral and Chemical Properties of the Yellow Fluorescent Protein zFP538 Chromophore† Ilia V. Yampolsky, Tamara A. Balashova, and Konstantin A. Lukyanov* Shemiakin-Ovchinnikov Institute of Bioorganic Chemistry, Miklukho-Maklaya 16/10, Moscow 117997, Russia Received April 27, 2009; Revised Manuscript Received July 17, 2009 ABSTRACT:

Members of the green fluorescent protein (GFP) family become chromophoric through a unique pathway based on autocatalytic modifications of their amino acid residues. The yellow fluorescent protein zFP538 from the button polyp Zoanthus possesses unique spectral characteristics that are intermediate between those of the green and orange-red fluorescent proteins. In this study, we used chemical synthesis to resolve conflicting data from crystallographic and biochemical analyses of the zFP538 chromophore structure. We synthesized 2-(5-amino-1-oxopentyl)-5-(4-hydroxybenzylidene)-3-methyl-3,5-dihydro-4Himidazol-4-one (5), which can spontaneously react intramolecularly to form cyclic imine (7). Compound 7 represents the native chromophore structure reported in the crystallographic study. We have also discovered an unusual isomerization of a 2-acylimidazolone to a 2,6-diketopiperazine derivative. The zFP538 chromophore is a complex system with intriguing chemical and spectral behavior, properties that have led to discrepancies in the interpretation of its structure. Our study supports the findings of previous crystallographic work, which postulated a cyclic imine chromophore structure within the native zFP538 protein, and also provides an explanation for experimental results obtained in the biochemical characterization of zFP538derived chromopeptides.

Members of the green fluorescent protein (GFP)1 family are now widely used as genetically encoded fluorescent labels (1). They enable the visualization of structures and processes in living cells and organisms. The practical use of GFP-like proteins is based on their unique ability to form chromophores inside protein globules via self-catalyzed reactions of post-translational modifications of internal amino acids (2). Chromophore maturation requires no additional enzymes or cofactors besides molecular oxygen. As a result, genes encoding GFP-like proteins can be introduced into nearly any cell or organism, leading to the functional expression of fluorescent proteins under the control of a specific promoter. GFP from the jellyfish Aequorea victoria was the first fluorescent protein characterized and cloned (3, 4). GFP homologues were subsequently found in corals (5), crustaceans (6), and even lower chordates (7). Coral GFP-like proteins are the most colorful homologues and include cyan, green, yellow, and red fluorescent proteins as well as nonfluorescent purple-blue chromoproteins (1, 8). Chromophore structures and the mechanisms of their formation in GFP-like proteins are under extensive study; these † This work was supported by Russian Foundation for Basic Research Grant 09-04-00356-a, the Molecular and Cell Biology program of the Russian Academy of Sciences, Howard Hughes Medical Institute Grant 55005618, and the program “State Support of the Leading Scientific Schools” (NS-2395.2008.4). K.A.L. is supported by grants from the President of the Russian Federation (MD-5815.2008.4 and MD2780.2009.4). *To whom correspondence should be addressed. Phone and fax: 7(495)3307056. E-mail: [email protected]. 1 Abbreviations: asFP595, purple chromoprotein from Anemonia sulcata; DMF, dimethyl formamide; DsRed, red fluorescent protein from Discosoma; GFP, green fluorescent protein from A. victoria; zFP538, yellow fluorescent protein from Zoanthus.

investigations provide an understanding of the “heart” of fluorescent proteins and suggest possible ways to modify their spectral properties. It has been demonstrated that the green chromophore in GFP is formed by cyclization of the protein backbone in the Ser65-Tyr66-Gly67 region (numbering of A. victoria GFP), followed by dehydrogenation of the CR-Cβ bond of Tyr66. As a result, a bicyclic structure of 5-(4-hydroxybenzylidene)-3,5-dihydro-4H-imidazol-4-one is formed: the six-membered aromatic ring of the Tyr66 side chain is linked to an unusual fivemembered heterocycle, which itself originates from condensation of the carbonyl carbon of Ser65 with the nitrogen of Gly67. Further chemical modifications of the green chromophore occur in red-shifted GFP-like proteins. In particular, oxidation of a CR-N bond results in an acylimine group conjugated to a GFP-like core in DsRed (9, 10). The DsRed-like chromophore is formed within many other proteins with red-shifted absorption and fluorescence (2). In some proteins, the acylimine moiety of the DsRed chromophore is further attacked by various nucleophiles to form additional types of red-shifted chromophores. For example, the chromophore in the purple chromoprotein asFP595 is formed by hydrolysis of the acylimine group, resulting in cleavage of the protein backbone and formation of a keto group conjugated to a GFP-like chromophore core (11, 12). In the orange fluorescent proteins mOrange and mKO, nucleophilic addition of Thr65 (in mOrange) or Cys65 (in mKO) side chain groups leads to unusual heterocycles without protein backbone scission (13, 14). The yellow fluorescent protein zFP538 from the button polyp Zoanthus demonstrates unique spectral characteristics that are intermediate between those of the green and orange-red fluorescent proteins (λmax,ex=528 nm, and λmax,em=538 nm) (5).

r 2009 American Chemical Society

Published on Web 07/18/2009

pubs.acs.org/Biochemistry

8078

Biochemistry, Vol. 48, No. 33, 2009

The chromophore-forming residues in zFP538 are Lys-Tyr-Gly. Differing opinions concerning the structure of the zFP538 chromophore have been published (15-17). In this study, we used chemical synthesis to resolve conflicting data from crystallographic (16, 17) and biochemical (15) analyses related to the zFP538 chromophore structure. This approach allowed us to establish a structural basis for the complex spectral transformations of the zFP538 chromophore described by Zagranichny et al. (15). EXPERIMENTAL PROCEDURES NMR spectra were recorded with Varian unity 600 MHz, Bruker Avance III 600 MHz, and Bruker DRX-500 MHz instruments. UV-vis spectra were recorded with a Varian Cary 100 spectrophotometer. Fluorescence excitation and emission spectra were recorded with a Varian Cary Eclipse fluorescence spectrophotometer. 2-(5-tert-Butyloxycarbonylaminopentyl)-5-(4-hydroxybenzylidene)-3-methyl-3,5-dihydro-4H-imidazol-4-one (3). N-Boc-6-aminohexanoylglycine (obtained by standard peptide synthesis) (5.77 g, 0.02 mol) was added to 100 mL of a THF solution containing dicyclohexylcarbodiimide (6.19 g, 0.03 mol). The reaction mixture was stirred overnight and then filtered and concentrated on a rotary evaporator. The resulting oil was dissolved in 50 mL of acetic anhydride, and sodium acetate (1.64 g, 0.02 mol) and p-acetoxybenzaldehyde (4.93 g, 0.03 mol) were added. The resulting mixture was stirred at 40 °C for 3 days until it solidified (indicating the formation of oxazolone 1). Most of the acetic anhydride was removed in vacuo, and ethanol (150 mL) was added followed by aqueous methylamine (c = 11.6 M, 8.6 mL, 0.1 mol). The mixture was vigorously stirred until the crystals of oxazolone 1 disappeared. Water (200 mL) was added, and crystals of 2 (with a crystalline admixture of dicyclohexylurea) were collected by suction and air-dried. The crude solid obtained was heated to reflux with 2 g of anhydrous cesium carbonate in 100 mL of DMF with vigorous stirring. The course of the reaction was monitored by TLC (SiO2, 10% chloroform/ethanol). The Rf of 2 was 0.33, while the Rf of 3 was 0.62. A spot-to-spot conversion was observed after 3 min. The reaction mixture was cooled, diluted with water (300 mL), and extracted with ethyl acetate (4  100 mL). The extracts were washed with brine, dried with anhydrous sodium sulfate, and evaporated. Column chromatography (SiO2, 10% chloroform/ethanol) gave pure 3 (4.96 g, 64%) as yellow crystals: 1H NMR (600 MHz, CDCl3) δ 1.43 (s, 9H, Boc), 1.46 (m, 2H, CH2), 1.57 (m, 2H, CH2), 1.83 (m, 2H, CH2), 2.59 (t, 2H, CH2), 3.16 (m, 5H, N-CH3 and CH2-NHBoc), 4.62 (br s, NH-Boc), 6.89 (d, 2H, aromatic), 7.08 (s, 1H, vinylic CH), 7.20 (br s, phenolic OH), 8.05 (d, 2H, aromatic); MS (ESI-MS) calcd for C21H30N3O4þ [M þ Hþ] 388.22, found 388.34. 2-(5-tert-Butyloxycarbonylamino-1-oxopentyl)-5-(4-hydroxybenzylidene)-3-methyl-3,5-dihydro-4H-imidazol-4one (4). Imidazolone 3 (1.14 g, 2.94 mmol) and SeO2 (0.46 g, 4.14 mmol) were placed in dioxane (60 mL) and refluxed for 8 h. The reaction mixture was filtered to remove Se, concentrated on a rotary evaporator, and chromatographed (SiO2, 10% chloroform/ethanol) to give crude 4 (720 mg). Recrystallization from methanol gave 4 as red needles (602 mg, 51%): 1H NMR (500 MHz, CD3OD) δ 1.42 (s, 9H, Boc), 1.57 (m, 2H, CH2), 1.74 (m, 2H, CH2), 3.10 (t, 2H, CH2), 3.15 (t, 2H, CH2),

Yampolsky et al. Table 1: 1H and 13C NMR Chemical Shifts in Compounds 5-7, Extracted from 1H, 13C, HMBC, HSQC, COSY, and TOCSY Spectra 5b atom numbera 1 2 3 4 5 6 7 8 9 10 11 12 13 14

1

H

6b 13

C

H

13

C

160.0 116.0 136.7 125.6 7.27 7.69 142.7 136.4 171.8 164.7 3.27 28.3 3.22 25.3 152.5 158.7 196.0 156.1 3.04-3.13 25.2 2.79 31.6 1.66-1.80 21.8, 26.3 1.77-1.83 21.6, 26.0 1.66-1.80 3.04-3.13 39.1 3.09 39.2 6.90 8.00

160.7 115.9 136.6 125.8 137.5

1

7c

6.95 8.09

1

H

13

C

6.61 8.02

161.7 120.3 137.2

7.21

135.5

3.43

2.76d 1.78 1.67 3.84

171.5 29.5 151.9 161.8 26.1 19.1 22.0 50.2

a Atom numbering according to Scheme 2. b A 3 mg/mL solution in a H2O/D2O mixture at pH 3.3 and 20 °C. c A 7 mg/mL solution in 2propanol-d8, saturated with Cs2CO3. d Integral intensity lowered ∼10-fold because of exchange with deuterium.

3.38 (s, 3H, N-CH3), 6.88 (d, 2H, aromatic), 7.35 (s, 1H, vinylic CH), 8.16 (d, 2H, aromatic); MS (ESI-MS) calcd for C21H28N3O5þ [M þ Hþ] 402.20, found 402.35. 2-(5-Amino-1-oxopentyl)-5-(4-hydroxybenzylidene)-3-methyl-3,5-dihydro-4H-imidazol-4-one Salt with Trifluoroacetic Acid (5). Boc-protected chromophore 4 (3 mg) was incubated with 0.2 mL of a 20% trifluoroacetic acid solution in dichloromethane for 30 min at room temperature. The volatiles were then removed on a rotary vacuum concentrator. Re-evaporation with dichloromethane was performed several times until no free trifluoroacetic acid remained. The solid obtained was pure by NMR and could be stored at -20 °C in a tightly closed vessel for 1 month with no detectable decomposition. NMR data are given in Table 1: MS (ESI-MS) calcd for C16H20N3O3þ [M þ Hþ] 302.15, found 302.28. RESULTS Synthesis. To obtain the target structure 5, we used an approach utilized in our previous work (12) that includes oxidation with SeO2 at a key stage (Scheme 1). We synthesized p-hydroxybenzylidene imidazolone 3, bearing a Boc-protected ω-aminopentyl substituent at the 2-position of the imidazolone, via a standard (18) set of transformations, including Erlenmeyer azlactonization, nucleophilic aminolysis of the azlactone, and cyclization of the dehydrotyrosine derivative 2 under basic conditions to form imidazolone 3. In this work, we have found optimal conditions for the aforementioned cyclization, namely, heating in DMF with Cs2CO3 as a base. These conditions provide quantitative yields, and reaction times are shortened to several minutes [we have tested these reaction conditions on several other substrates (data not shown)]. Previously, this transformation was accomplished by prolonged (several hours) heating in aqueous or alcoholic media with alkaline metal carbonates or hydroxides (12, 18, 19). When substituents at positions 1 and 2 of the imidazolone are simple alkyls, both methods give good results, but the latter method is strongly preferred when substituents are sterically hindered or when competing side reactions can take place.

Article

Biochemistry, Vol. 48, No. 33, 2009

8079

Scheme 1: Synthesis of the zFP538 Chromophore (in noncyclized form 5)a

a (i) (a) Glycine ethyl ester hydrochloride, ethanolic NaOH, DCC; (b) aqueous NaOH, ethanol, 50 °C, then HCl (78%); (ii) DCC, then p-acetoxybenzaldehyde, NaOAc; (iii) MeNH2; (iv) Cs2CO3, DMF, reflux for 5 min (64% over three stages); (v) SeO2, dioxane, reflux (51%); (vi) DCM, TFA, room temperature, 30 min (100%).

Scheme 2: pH-Dependent Chemical Transformations of the zFP538 Chromophore

Imidazolone 3 was oxidized to the Boc-protected zFP538 chromophore precursor 4 by being heated in a dilute dioxane solution with excess SeO2. Finally, 4 was smoothly deprotected with trifluoroacetic acid in dichloromethane at room temperature to give 5 as the TFA salt. Chemical and Spectral Behavior. Although the structure of compound 5 does not correspond to the structure of the mature chromophore within zFP538, we expected that 5 would intramolecularly cyclize to form cyclic imine 7 (Scheme 2). Indeed, we observed this transformation upon basification of the solutions of trifluoroacetate of 5 (pH increased from 5 to 8.5) in water or in 2-propanol. NMR studies clearly indicated the presence of linear aminoketone 5 in acidic D2O and 2-propanol-d8. In particular, the 13C chemical shift of C-10 (Scheme 2) was 196.0 ppm (D2O), which is characteristic of a carbonyl group, and no cross-peak

was detected in the C-H HMBC spectrum between CH2-14 and C-10. In D2O at pH 8.7, the 1H NMR spectrum was complex, containing several sets of signals (not shown). Notably, this complexity of the spectrum was not a result of chemical degradation of 5, since the reappearance of the initial, single set of signals was observed upon acidification to pH 5. This complex behavior (which probably reflects reversible hemiaminal formation in water) impeded unambiguous structure determination. To overcome this problem, we used 2-propanol-d8 with anhydrous Cs2CO3 as a base. In this case, we observed a single set of signals in the NMR spectra corresponding to cyclic imine 7. In particular, the 13C chemical shift of C-10 became 161.8 ppm, which is characteristic of an imine, and a cross-peak between CH2-14 and C-10 appeared in the C-H HMBC spectrum. Interestingly, the integral intensity of the signal from CH2-11 in the 1H spectrum

8080

Biochemistry, Vol. 48, No. 33, 2009

Yampolsky et al.

FIGURE 1: Absorption (A-C) and fluorescence (D-F) spectra of model compounds 5-7 in water solutions. (A) Absorption spectra of 5 at pH 4.0

(---) and 7 at pH 8.9 (;). (B) Change in the absorption spectrum upon basification of the solution of 5 to pH 8.9. Spectra were recorded at 10 s intervals and show the transformation of a short-lived intermediate (deprotonated 5), absorbing at ∼530 nm, into 7, absorbing at 468 nm. (C) Absorption spectra of 6 at pH 3.0 (---) and pH 8.0 (;). (D) Excitation and emission spectra of 5 at pH 4.0 (gray lines) and 7 at pH 8.9 (black line). (E) Excitation and emission spectra of the short-lived intermediate (deprotonated 5) obtained upon basification of the solution of 5 to pH 8.9. (F) Excitation and emission spectra of 6 at pH 3.0 (gray lines) and pH 8.0 (black lines). In panels D and E, within each pair of lines the emission spectrum is the one at longer wavelengths.

decreased ∼10-fold. This effect can be attributed to H-D exchange due to imine-enamine tautomerization. At pH